Explore California Housing Dataset

 

California Housing Dataset

---

🔹 Dataset Characteristics

Feature	Description
Number of Instances	20,640
Number of Attributes	8 numerical predictive attributes + 1 target + 1 categorical
Target Variable	Median House Value

---

🔹 Context

This dataset is used in the book
“Hands-On Machine Learning with Scikit-Learn and TensorFlow” by Aurélien Géron.

It is widely used as an introductory dataset for machine learning because:

• Requires basic preprocessing

• Has clear and interpretable features

• Is moderate in size (not too small, not too large)

---

🔹 Dataset Description

The dataset contains information about housing in California districts based on the 1990 U.S. Census.

Each row represents a census block group, which is:

• The smallest geographical unit used by the census

• Typically contains 600 to 3,000 people

---

🔹 Features (Attributes)

No.	Attribute	Description
1	longitude	How far west a house is (higher = farther west)
2	latitude	How far north a house is (higher = farther north)
3	housing_median_age	Median age of houses in a block
4	total_rooms	Total number of rooms in a block
5	total_bedrooms	Total number of bedrooms in a block
6	population	Total population in the block
7	households	Number of households in the block
8	median_income	Median income (in tens of thousands of USD)
9	median_house_value	Median house value (target variable)
10	ocean_proximity	Categorical feature (distance from ocean)

---

🎯 Target Variable

• median_house_value

• Represents housing prices in hundreds of thousands of dollars

  • Example: value = 2.5 → $250,000

---

⚠️ Important Notes

• Dataset is not fully cleaned

• May require:

  • Handling missing values (e.g., total_bedrooms)

  • Encoding categorical variable (ocean_proximity)

  • Feature scaling

---

🔍 Additional Insights

• A household = group of people living in one home

• Some blocks may have:

  • Few households

  • Many rooms (e.g., vacation areas)

• This can lead to unusual feature values

---

📥 Dataset File

• File name: housing.csv

• Source: California Housing Dataset (1990 Census)

• Download File from kaggle : housing.csv

Experiment-1

1.Download the file housing.csv

2.Open colab

3.Start a new python Notebook

4.Upload the housing.csv file

5.Analyse the data file

6.Visualize Data

Python Code

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

# Reading the data file

df = pd.read_csv("/content/sample_data/housing.csv")

#print the datframe

print(df)

#printing the shape

print(df.shape)

#printing top 10 and bottom 10 rows

print(df.head(10))

print(df.tail(10))

#printing the columns

print(df.columns)

#print a feature value

print(df['total_rooms'])

# Selecting one input feature and target feature (dependent) for 150 rows only

X = df['total_rooms'].head(150).values.reshape(-1,1)

y = df['median_house_value'].head(150).values.reshape(-1, 1)

#visualizing these values

plt.scatter(X, y)

plt.xlabel("Total Rooms")

plt.ylabel("Median House Value")

plt.title("Scatter Plot of Total Rooms vs. Median House Value")

plt.show()

#printing the statistical summary

print(df.describe())

Experiment-2

learn to use toy data sets from scikit-learn

Caifornia hosuing dataset can be fetched and used directly with scikit-learn ( understand how to use it )

Scikit-learn (sklearn) provides a wide range of sample datasets specifically designed for practicing machine learning tasks without needing to find or clean outside data. These datasets are categorized into three main types: Toy datasets, Real-world datasets, and Generated (Synthetic) datasets

from sklearn.datasets import fetch_california_housing

import pandas as pd

import matplotlib.pyplot as plt

housing = fetch_california_housing()

print("Printing the Descriptions of dataset")

print(housing.DESCR)

print("Printing the shape of data")

print(housing.data.shape)

print("Printing the shape of target data")

print(housing.target.shape)

print("Printing the feature names")

print(housing.feature_names)

print("Printing the target names")

print(housing.target_names)

df = pd.DataFrame(housing.data, columns=housing.feature_names)

df['MedHouseVal'] = housing.target

print("\nShape of Data:")

print(df.shape)

print("\nFirst 10 Rows:")

print(df.head(10))

print("\nLast 10 Rows:")

print(df.tail(10))

print("\nColumns:")

print(df.columns)

print("\nStatistical Summary:")

print(df.describe())

print("\nMissing Values:")

print(df.isnull().sum())

# -----------------------------

# 3. Scatter Plot (Rooms vs Price)

# -----------------------------

plt.figure()

plt.scatter(df['AveRooms'], df['MedHouseVal'])

plt.xlabel("Average Rooms")

plt.ylabel("House Value")

plt.title("Rooms vs House Value")

plt.show()